Machine learning-assisted development of polypyrrole-grafted yarns for e-textiles

Abstract

The development of digitally enhanced fabrics is growing, but progress is currently being hampered by a lack of sustainable alternatives to metallic conductors. In particular, the process of testing and optimizing new candidate materials is both time-consuming and resource intensive.To address these challenges, we present a machine learning-assisted approach to the design of fully-textile based conductive e-textile prototypes. Based on commercially available Tencel yarn coated with polypyrrole, with 11 experiments we were able to establish the global optimum of the reaction and estimate the noise, crucial for the understanding of the electrical resistance’s behavior. The reaction conditions are optimized for conductivity and cost-effectiveness by means of Bayesian optimization and Pareto front analysis. Notably, we find that the addition of p-toluenesulfonic acid as a dopant does not significantly influence the conductivity of the yarn and provide a possible rationale based on the surface morphology of the yarn. The optimized yarns are woven into prototype fabrics with different patterns, and we demonstrate their applicability as flexible conductive wearable and heaters.

Keywords

• E-textiles
• Polypyrrole
• Conductive yarns
• Machine learning
• Bayesian optimization
• Cost evaluation